Create flashcards free and with AI

Karpicke and Roediger (2008, Science 319:966 to 968, doi:10.1126/science.1152408) studied retrieval practice in a controlled experiment. Quanta uses the question-and-answer format for active recall; the study is methodological context, not evidence of Quanta product effectiveness.

Quanta uses FSRS-6 (Free Spaced Repetition Scheduler, version 6). In the open comparison run by the open-spaced-repetition community, FSRS-6 reaches a log-loss of 0.3460 on 349,923,850 reviews from 9,999 collections (retrieved on 5 September 2026); log-loss measures prediction error, lower is better, and that table carries no row for SM-2. The peer-reviewed paper behind it, Ye et al. 2022, reports 220 million behaviour logs and a 12.6% improvement over the state of the art. Both are results of those datasets, not a blanket product effect. Anki now supports FSRS too; Quanta integrates the scheduler without separate setup. New cards receive their next FSRS date after the first user rating.

Distractor check: the Quanta prompt requires a plausibility check of multiple-choice wrong answers before output. Basis: Haladyna, T. M. and Downing, S. M. (1989), Applied Measurement in Education 2(1), 37 to 50, doi:10.1207/s15324818ame0201_3. The prompt rule does not guarantee error-free distractors.

The Quanta AI tutor asks contextual follow-up questions and can then provide feedback and a model answer. A separate per-card AI explanation supplies a direct contextual explanation when requested. Chi, M.T.H. et al. (2001), Learning from human tutoring, Cognitive Science 25(4), 471 to 533, doi:10.1207/s15516709cog2504_1, studies human tutoring and is a pedagogical reference here, not proof of product effectiveness.

Method references: Karpicke and Roediger (2008) report about 80% of the vocabulary pairs recalled after one week in the two conditions with repeated retrieval practice, and 36% and 33% in the two conditions where pairs were dropped from further testing once recalled; those figures describe that experiment and are not a general promise for every learning situation or for Quanta. In the open comparison run by the open-spaced-repetition community, FSRS-6 reaches a log-loss of 0.3460 on 349,923,850 reviews from 9,999 collections (retrieved on 5 September 2026); log-loss measures prediction error, lower is better, and that table carries no row for SM-2. The peer-reviewed paper behind it, Ye et al. 2022, reports 220 million behaviour logs and a 12.6% improvement over the state of the art. Both are results of those datasets, not a blanket product effect. Anderson and Krathwohl (2001) describe Bloom taxonomy; Haladyna and Downing (1989) describe MC item rules. These sources support the respective methods, not an automatic learning-outcome guarantee for Quanta.